"Private search engine" is not one claim. It is at least three, they are independent of each other, and an engine can satisfy any one of them while failing the others. Reading this category usefully means separating them before you look at a single logo.
The three questions that "private" can answer
- Does the operator record what you searched and link it to you? This is the promise most privacy engines lead with: no query logs tied to an IP address or an account, no behavioural profile, no personalised results, no tracking cookies. It is a promise about the engine's own conduct, and it is the easiest of the three to make and the hardest for an outsider to verify — nobody outside the company can audit a log file that is claimed not to exist.
- Does your query leave the engine at all? This is a structural question with a factual answer, and it is the one this site puts first. If an engine has no index of its own, then every search you make is forwarded to somebody else's servers. That is not automatically bad — the forwarding can be done from the engine's own infrastructure, so the upstream provider sees the query but not you. But it means a third party sees the words you typed, and that third party is usually Microsoft or Google.
- Whose law applies? Where the company is incorporated and where the servers sit determine who can compel what. A European engine is inside the GDPR enforcement structure. A Swiss engine is not in the EU. A UK engine is subject to the Investigatory Powers Act 2016. A US-owned engine is exposed to US legal process regardless of how European its branding is. The usual defence — offered by more than one engine here, and a reasonable one — is that data that was never retained cannot be handed over.
Most confusion in this category comes from collapsing questions one and two. "We do not track you" and "your query never reaches Google" are different statements, and several well-regarded engines make the first while explicitly not making the second.
Why most privacy engines run on somebody else's index
The reason is economic and it is not a secret. Crawling the open web continuously, storing it, and ranking it is a permanent heavy cost that scales with the size of the web rather than with the size of your audience. A privacy-focused engine typically has a small audience by construction, refuses behavioural advertising by principle, and therefore has the least revenue per query of anyone in search. Buying a results feed converts that impossible fixed cost into a variable one and lets a small team ship a real product. That is why the category is full of front ends.
Two events have squeezed that model. Microsoft raised Bing API pricing sharply in 2023, and retired the public Bing Search APIs altogether on 11 August 2025, replacing them with an AI-grounding service rather than a results interface. Engines with longstanding direct syndication contracts were not immediately affected; engines relying on the public API were. The visible response has been a push toward European index-building — Ecosia and Qwant announced a 50:50 joint venture, European Search Perspective, in November 2024 and began serving live traffic from its Staan index in early August 2025, starting in France. Both companies have published targets for how much of their traffic that index should carry; neither has published what it has actually achieved, and targets are not results.
What the nine engines here actually are
The spread within this single category is wider than between most categories on the site.
- Genuinely own-index. Mojeek has crawled and ranked its own index since the mid-2000s and has never resold Google's or Bing's results at any point — it is the least ambiguous case in the group, and also one of the smallest indexes, at a published figure of around nine billion pages in 2025 against a Google index generally understood to be in the hundreds of billions. Brave Search launched in 2021 taking roughly an eighth of its results from a third-party API and announced the removal of the last of those calls in April 2023, adding its own image and video indexes that August. Marginalia is a working independent crawler and index built and run by one person, Viktor Löfgren, with grant funding rather than a business model, and is deliberately tuned toward non-commercial pages that mainstream ranking buries.
- Hybrid. DuckDuckGo runs its own crawlers and its own indexes — those power the instant answers above the links — but says on its own help pages that the traditional links and images are largely sourced from Bing. Both of the common summaries of DuckDuckGo are wrong: it is not an independent index, and it is not merely a Bing skin. Qwant crawls at genuine scale and supplements with Bing where it judges its own relevance insufficient, and for images.
- Relays and front ends. Startpage has no crawler and no index by design: it submits your query to third parties from its own servers, strips the identifying data, and returns their results. It was Google-only from July 2009 until March 2023, when Bing web results were added; its documentation names both today, and the split between them is not published. Swisscows describes an exclusive Bing partnership on its own blog. Gibiru publishes no crawler, no user-agent and no index-size figure; its founder described it in 2012 as a modified Google feed, and sources disagree about whether an index exists at all.
Ownership is part of the privacy question
Who owns a privacy engine is a material fact about it, and the answers are not uniform. Ecosia's founder transferred his shares into a steward-ownership structure with the Purpose Foundation in October 2018, permanently surrendering the ability to sell the company or take dividends out of it — an unusually strong structural guarantee that survives any change of management. Startpage, by contrast, has been majority-owned since late 2018 by Privacy One Group, a subsidiary of the US performance-advertising company System1, a fact that was not disclosed at the time and caused a well-documented controversy when it emerged. Mojeek is a small independent English company. Marginalia has no company at all. None of these facts settles whether an engine keeps its promises, and all of them belong in the assessment.
The practical upshot: if your concern is advertising profiles and personalised results, most engines in this category address it and the choice between them is largely about interface and coverage. If your concern is that Google or Microsoft should not see your queries at all, the list of engines that satisfy you is short, and it is exactly the list with its own index.